from plasem_taln import filters_baseline_similarity
similarity = filters_baseline_similarity
methodname = 'baseline_filters'
caption = 'Semeval - Scores MAP - Méthodes de référence avec filtres'
parameters = list(product(corpora, filters_partition))
parameters_description = ('Édition', 'Filtres', 'Score MAP')
description_functions = [lambda x: x, get_filters_descr]
for corpus, *rest in parameters:
context['filters'] = [filters[key] for key in rest[0]]
from plasem_semeval import write_scores_to_file
from plasem_taln import comparator
comp = comparator(context, similarity)
scores = make_score_tree(doctrees[corpus], comp.getscore)
predfile = getpredfilename(methodname, corpus, *rest)
write_scores_to_file(scores, predfile)
from plasem_semeval import sorted_scores_from_semeval_relevancy
from plasem_algostruct import mean_average_precision
MAP = mean_average_precision(
sorted_scores_from_semeval_relevancy(relevancy[corpus],
scores).values())
restable.append([
*(description_functions[i](value)
for i, value in enumerate((corpus, *rest))),
'%.2f' % (100 * MAP)
])
result = {
org: average_precision(values)
for org, values in sscores.items()
}
return result
relevdict = {
corpus: relevancy_dict_relevancy(relevancy[corpus])
for corpus in corpora
}
for corpus, *rest in parameters:
context['filters'] = [filters[key] for key in rest[0]]
compfilters = comparator(context, filters_baseline_similarity)
scores = make_score_tree(doctrees[corpus], compfilters.getscore)
predfile = getpredfilename(methodname, corpus, *rest)
write_scores_to_file(scores, predfile)
MAP = mean_average_precision(
sorted_scores_from_semeval_relevancy(relevancy[corpus],
scores).values())
if MAP > best[corpus]:
best[corpus] = MAP
bestmeth[corpus] = rest[0]
bestaps[corpus] = avgtree(scores, relevancy[corpus])
bestscores[corpus] = scores
# restable.append([*(description_functions[i](value)
# for i, value in enumerate((corpus, *rest))),